What the ISACA AAISM is and who it is for
The Advanced in AI Security Management certification is ISACA’s first AI centric security management credential. It is built for experienced security professionals who need to manage, govern and secure AI systems in enterprise environments. It sits above the traditional security management certifications rather than beside them: the question is not whether you understand security, but whether your security judgement holds up when the system in front of you is an AI model.
That is reflected in the entry rule. You need an active CISM or CISSP to register, so everyone sitting the exam already has a recognised security management or security professional certification. Once you register, you have a six month eligibility window in which to sit the exam.
It suits security managers, architects and consultants whose organisations are adopting AI and who have been asked to put controls, governance and risk management around it.
ISACA AAISM at a glance
| Item | Detail |
|---|---|
| Exam code | AAISM |
| Questions | 90 |
| Time allowed | 150 minutes |
| Exam fee | US$459 for ISACA members, US$599 for non-members |
| Where you sit it | PSI testing centres or remote proctoring |
What is on the exam
Three domains. AI technologies and controls is the largest at 38%, and the two management domains take 31% each, so the exam leans slightly towards the technical side while still giving management topics most of the marks.
AI Technologies and Controls (38%). The technical core. It covers how AI systems are built and run, the threats specific to them, and the controls that protect them. Expect scenarios involving model attacks, data poisoning, adversarial inputs, prompt manipulation, leakage of training data and weaknesses in the AI supply chain, and questions that ask which control, monitoring approach or architectural choice best reduces a given risk.
AI Governance and Program Management (31%). How an organisation governs AI as a programme rather than as a series of projects. That means policies and standards for AI use, roles and accountability, oversight of AI across its life cycle, handling third party and vendor AI, awareness and training, and fitting AI into the existing security programme. ISACA frames many of these questions from the point of view of a manager deciding what should happen first or who should own a decision.
AI Risk and Opportunity Management (31%). Identifying, assessing and treating AI risk, and weighing it against the business value AI offers. Topics include AI risk assessment, risk appetite and tolerance, impact on privacy and ethics, monitoring and reporting AI risk, and deciding when the opportunity justifies the exposure. The exam expects you to enable the business safely, not to block AI by default.
Why people fail it
Most candidates already hold a CISM or CISSP and know how to protect traditional IT systems. That is exactly the trap. The AAISM tests whether that judgement extends to AI, and the risks are different in kind: a model can be poisoned through its training data, manipulated through its inputs or made to reveal data it was trained on, and none of those map neatly onto the controls you use for a web server or a database. Candidates who answer from their existing security instincts pick controls that sound right but do not address the AI specific risk in the scenario.
The second trap is the style of question. The AAISM is scenario based and tests applied judgement rather than memorisation. Like other ISACA exams, it often offers several defensible answers and asks for the best, first or most important one from a management perspective. Technical candidates tend to jump to the most hands on fix, when the answer being marked is to establish governance, assess the risk or involve the right owner first.
Finally, the balance of the domains catches people out. AI technologies and controls is the largest single domain, but the two management domains together carry the majority of the marks. Candidates who focus only on the technical AI threats, because that is the newest and most interesting material, under prepare for the governance and risk questions that make up most of the exam.
A study plan that fits the exam
Six weeks, weighted to the domains: two for AI technologies and controls, one each for governance and risk, then a week of mixed scenarios and a week of timed practice.
- Week 1: AI technologies and controls, part one. How AI systems are built, trained and deployed, and the threats that target them: poisoning, adversarial inputs, prompt manipulation, data leakage and supply chain risk. At the end of the week, try the free AAISM practice questions to see how ISACA frames its scenarios.
- Week 2: AI technologies and controls, part two. The controls: access management, monitoring of model behaviour, validation, secure architecture and testing. For each threat from week one, be able to name the control that best addresses it and explain why the alternatives fall short.
- Week 3: AI governance and program management. Policies, accountability, life cycle oversight, third party AI and fitting AI into the wider security programme. Practise answering from a manager’s point of view: what comes first and who owns it.
- Week 4: AI risk and opportunity management. Risk assessment, appetite and tolerance, privacy and ethical impact, and reporting. Work the matching questions in the AAISM practice question pack and read the reasoning for every wrong option, which is where ISACA’s framing becomes clear.
- Week 5: mixed scenarios. Work sets that mix all three domains, since the exam will not tell you which domain a question belongs to. For every miss, note whether you misread the risk or chose a technical fix over a management answer.
- Week 6: timed runs. Use the questions only PDF for full sittings against the 150 minute limit, score by domain, and spend the last days on the weakest one.
On exam day
You sit the AAISM at a PSI testing centre or through remote proctoring. For remote delivery, check the requirements and run any system check well before the day. Remember that your eligibility window runs for six months from registration, so book a date inside it.
You have 150 minutes for 90 questions, which gives you time to read each scenario properly. Use it. Many wrong answers come from missing a single word such as first, best or most, so read the question stem twice before you look at the options. ISACA does not publish a passing score for AAISM, so there is no point trying to estimate your result as you go. Answer every question before time runs out.
Frequently asked questions
Can I sit the AAISM without a CISM or CISSP?
No. You need an active CISM or CISSP to register.
What happens if I fail? Can I retake it?
You can, but each attempt costs the full fee again: US$459 for ISACA members and US$599 for non-members. ISACA publishes its retake rules, including any limits and waiting periods, on its website, so check them before you rebook. The practice pack is refunded if you fail, but the exam fee is not.
Is the practice question pack enough on its own?
No. The pack is 371 practice questions mapped to the AAISM objectives, with the reasoning behind every answer and every wrong option, covering AI governance, risk management and technical controls. It is practice: it shows you how ISACA frames AI security scenarios and exposes the areas where your existing security instincts lead you astray. It is not a course, so use it alongside the official ISACA material.
When you are ready to test your judgement, get the 371 question AAISM pack for US$39, pass or your money back.